# Cursor Analytics MCP for AI Agents MCP > Cursor Analytics MCP lets you pull team-wide coding metrics directly into your AI agent. Track things like AI edit acceptance rates, autocomplete stats, and model usage without leaving your chat interface. It turns your analytics data into conversational insights for managers and leads. ## Overview - **Category:** developer-tools - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_H9OWwxFXSPPU6eyXPduRCKXQ8tGqPSGu6oKasFrL/mcp - **Tags:** team-analytics, ai-metrics, usage-insights, enterprise, coding-metrics ## Description It's a headache for engineering leads to figure out if the team is actually getting value from AI coding tools. You usually have to jump between different dashboards, export CSVs, and manually crunch numbers to see if everyone is actually using the right models or if they're just clicking accept on every suggestion without thinking. This MCP changes that by bringing those metrics directly into your natural conversation with your AI agent. Instead of hunting for data, you can just ask about your team's adoption trends or see who's leading the leaderboard. It pulls in everything from tab autocomplete stats to specific MCP tool usage, making it easy to spot where people are struggling or where the tools are hitting the mark. By connecting this through the Vinkius marketplace, you get a unified way to keep your pulse on engineering productivity without the manual data gathering. It's about moving from I think they're using it to I know they're using it in seconds. ## Tools ### get_agent_edits Get metrics on suggested code edits, including acceptance rates and lines changed. ### get_plans_adoption See how often your team uses the plan mode to outline code before writing it. ### get_ask_mode_adoption Check how frequently users use ask mode to query their codebase. ### get_client_versions View the distribution of Cursor client versions across your organization. ### get_commands_adoption See which specific Cursor commands are most popular among your developers. ### get_daily_active_users Get the count of daily active users, including breakdowns by CLI and cloud activity. ### get_leaderboard Pull a ranked list of team members based on various AI usage metrics. ### get_mcp_adoption Track which MCP tools your team integrates most often. ### get_model_usage See which AI models your team prefers based on message volume and active users. ### get_skills_adoption Monitor which specific Cursor skills are being leveraged the most. ### get_tab_usage Get data on tab autocomplete suggestions, accepts, and rejects. ### get_top_file_extensions Identify the top 5 file extensions ranked by suggestion volume each day. ## Prompt Examples **Prompt:** ``` What's the AI model usage like for the team this week? ``` **Response:** ``` ### Model Usage Breakdown Over the last 7 days, here is how the team is utilizing different models: | Model | Messages Sent | Active Users | | :--- | :--- | :--- | | **claude-sonnet-4** | 4,200 | 12 | | **gpt-4** | 1,800 | 8 | | **gemini-2.5-pro** | 900 | 5 | **Key Insight:** Claude Sonnet 4 currently accounts for 60% of all team interactions. ``` **Prompt:** ``` Show me the edit acceptance rate for the last 30 days. ``` **Response:** ``` ### Agent Edit Metrics (Last 30 Days) * **Total Suggested Diffs:** 15,000 * **Total Accepted Diffs:** 9,200 * **Overall Acceptance Rate:** **61%** **Trend Analysis:** * The team accepted 28,500 of 42,000 suggested lines (68% line acceptance). * Acceptance rates have been trending upward, moving from 55% in week one to 67% in week four. ``` **Prompt:** ``` Who are the top users of the AI tools? ``` **Response:** ``` ### Team AI Usage Leaderboard Based on the combined lines accepted and active interactions, here are the top contributors: 1. **Sarah Chen** - 4,200 lines accepted 2. **Marcus Thorne** - 3,850 lines accepted 3. **Elena Rodriguez** - 3,100 lines accepted 4. **David Wu** - 2,900 lines accepted *Would you like to see the full leaderboard with more metrics?* ``` ## Capabilities ### Track AI edit acceptance See how many suggested code changes your team actually keeps. ### Monitor tab autocomplete Get data on how much your team relies on inline code completions. ### View model usage See which specific AI models your developers prefer to use. ### Monitor active users Track daily active user counts across different workflows. ### Identify top file types Discover which languages your team codes in most based on suggestions. ### Rank team contributors Generate a leaderboard based on various AI interaction metrics. ## Use Cases ### The ROI Report An Engineering Manager asks for a monthly summary of AI edit acceptance rates to justify the enterprise seat cost. ### The Tool Audit A Platform Team wants to see which MCP tools are actually being used to decide which ones to keep. ### The Training Plan A Team Lead notices low plan mode adoption and asks for a breakdown to target a specific team for a workshop. ### The Cost Check A DevOps Engineer wants to see which models are most popular to optimize the team's spend. ## Benefits - Stop manual data gathering by pulling team metrics directly into your chat. - Measure actual ROI on AI tools using get_agent_edits and get_tab_usage. - Identify training gaps by seeing which models or skills are underused via get_model_usage. - Plan smoother rollouts by checking get_client_versions to spot outdated software. - Spot high-performing contributors quickly with the get_leaderboard tool. - Understand your tech stack better by tracking get_top_file_extensions daily. ## How It Works The bottom line is you get conversational access to your team's AI coding performance data. 1. Subscribe to the Cursor Analytics MCP on Vinkius. 2. Enter your Cursor Analytics API key with admin:* scope. 3. Ask your AI agent for specific team metrics or adoption trends. ## Frequently Asked Questions **How do I get a Cursor Analytics API key?** Sign in to your Cursor Dashboard, navigate to Settings → API Keys, and generate a new API key. Ensure the key has the `admin:*` scope — Analytics API access requires it. The key format is `crsr_xxx`. Copy it immediately as it may not be shown again. Paste it into the API key field above to connect. **Is an Enterprise plan required to use the Analytics API?** Yes. The Cursor Analytics API is available exclusively to Enterprise teams. If your team is not on an Enterprise plan, API requests will return HTTP 403 Forbidden. Contact your Cursor account manager or upgrade to Enterprise to gain access. **What analytics metrics are available?** This MCP provides 12 analytics endpoints: agent edits (suggested/accepted/rejected diffs and lines), tab autocomplete usage (suggestions, accepts, rejects, lines), daily active users (with CLI, cloud agent, and bugbot breakdowns), client version distribution, AI model usage (messages and users per model per day), top 5 file extensions per day, MCP tool adoption, Cursor command adoption, plan mode adoption, skills adoption, ask mode adoption, and a team leaderboard with configurable sorting and pagination. **What date formats does the API support?** All analytics endpoints support the following date formats for startDate and endDate: relative shortcuts (`7d`, `30d`, `today`, `yesterday`, `now`), ISO 8601 strings (e.g. `2026-07-10T00:00:00Z`), calendar dates in `YYYY-MM-DD` format, and Unix timestamps. When omitted, startDate defaults to `7d` and endDate defaults to `today`. **Can I filter analytics by specific users?** Yes. All analytics endpoints accept an optional `users` parameter with comma-separated email addresses or user IDs. When provided, the API returns metrics filtered to only those users. This is useful for tracking individual or subgroup performance, or comparing team members on the leaderboard. **What are the API rate limits?** The Cursor Analytics API enforces the following rate limits: 100 requests per minute at the team level for standard analytics endpoints, 50 requests per minute for by-user analytics, and 20 requests per minute for conversation insights endpoints. If you exceed these limits, the API returns HTTP 429 and the engine surfaces a clear message asking you to wait. **Does the API support caching with ETags?** Yes. The Cursor Analytics API supports ETag-based caching. When you make a request, the API may return an `ETag` header. On subsequent requests, sending `If-None-Match` with that ETag value will return HTTP 304 Not Modified if the data hasn't changed, saving bandwidth and reducing rate limit consumption. The engine handles standard HTTP responses transparently.